June 2026 Cyber Attacks and Data Breaches: A Roundup With a Severity Read

June 2026 was a month of identity-driven data theft and software-supply-chain compromise far more than classic ransomware. A critical Oracle PeopleSoft zero-day drove a mass breach of universities, Microsoft shipped its largest-ever Patch Tuesday with an actively-exploited Defender flaw inside it, a credential-stealing worm poisoned dozens of Red Hat npm releases, and large breaches landed at iRhythm, Novo Nordisk, Tata Electronics and Kodak. This is an independent, attributed digest: every figure is sourced, each item is marked confirmed, claimed or reported, with a defensive takeaway. PlayCISO is independent and not affiliated with any organisation named here; this is general information, not commentary on any specific party’s conduct.

Method, and a note on the severity scores

We checked each item against primary reporting, vendor statements and CISA advisories, corrected attributions where sources disagreed, and flagged anything that rested only on an attacker’s word. Two examples of corrections worth calling out: the Novo Nordisk and Tata Electronics intrusions were attributed by some trackers to “Hunters International,” but primary reporting names FulcrumSec and World Leaks respectively — and the mid-June Iranian bank disruptions were widely pinned on Predatory Sparrow, which actually matches that group’s June 2025 Bank Sepah attack, not a confirmed 2026 claim.

The editorial severity beside each entry is a subjective PlayCISO reading out of ten — how much we think a security leader should care, given scale, sensitivity and how actionable the fix is. It is not CVSS, not EPSS and not any recognised standard; use it as a quick relative signal only. Several of the highest-profile numbers this month are unverified threat-actor claims, not confirmed facts.

Confirmed: actively-exploited vulnerabilities and patches

Oracle PeopleSoft zero-day (CVE-2026-35273). Status: Confirmed. Editorial severity: 9/10. A missing-authentication flaw in PeopleSoft PeopleTools (the Environment Management component behind the PSEMHUB hub, affecting versions 8.61 and 8.62), rated CVSS 9.8 and exploitable over the network with no credentials. Google’s Mandiant attributes exploitation between roughly 27 May and 9 June to the extortion group ShinyHunters (tracked as UNC6240), and reports put the impact at more than 300 instances across 100-plus organisations, hit hardest among universities; the University of Nottingham was among the first to confirm. Oracle issued an out-of-band advisory on 10 June, after exploitation had begun. Takeaway: inventory every internet-reachable PeopleSoft instance, apply Oracle’s mitigations and patches, and hunt for prior exploitation — an out-of-band advisory means attackers were already inside.

Microsoft June Patch Tuesday — a record haul with an exploited Defender flaw. Status: Confirmed. Editorial severity: 7/10. Microsoft’s 9 June update was its largest on record — reported at around 206 CVEs, though Tenable counted 198, so the sources disagree on the total — and included CVE-2026-41091, a Microsoft Defender elevation-of-privilege flaw exploited in the wild and fixed through Defender’s automatic updates. (A separate, hyped flaw dubbed “RoguePlanet,” CVE-2026-50656, was disclosed later in the month amid a researcher dispute and patched in July — it was not part of this release.) Takeaway: do not try to swallow 200-plus fixes at once — rank by “actively exploited / public proof-of-concept,” confirm the Defender engine updated, and patch the exploited item first.

• SimpleHelp (CVE-2026-48558). A CVSS 10.0 authentication-bypass flaw in the SimpleHelp remote-support tool, added to the CISA KEV catalog and tied to active use by malware tracked as TaskWeaver and Djinn Stealer. Status: Confirmed. Editorial severity: 8/10. Takeaway: remote-access and RMM tooling is a top ransomware entry point — patch it and restrict it to known management networks.
• Ivanti Sentry (CVE-2026-10520) and Check Point Security Gateway (CVE-2026-50751). Both landed in CISA KEV in June: an Ivanti Sentry OS-command-injection flaw rated CVSS 10.0, and a Check Point gateway authentication flaw (CVSS 9.3) linked to exploitation by the Qilin ransomware operation. Separately, Belgium’s state security service was reported breached via an Ivanti flaw, with suspected China-linked actors. Status: Confirmed (KEV). Editorial severity: 8/10. Takeaway: edge and VPN appliances remain the perimeter attackers prize — patch on day one and watch for post-exploitation.
• 23 vulnerabilities added to CISA KEV in June. Beyond the above, confirmed-exploited additions spanned Splunk Enterprise, three CVSS 10.0 Ubiquiti UniFi OS flaws, Cisco SD-WAN Manager, Google Chrome’s V8 engine and PTC Windchill. Status: Confirmed. Editorial severity: 7/10. Takeaway: the KEV catalog is your free, evidence-based patch queue.

Confirmed: supply-chain compromises

• Red Hat npm packages — a Shai-Hulud-style worm. On 1 June, Wiz found dozens of package releases under the @redhat-cloud-services npm namespace infected with a self-propagating credential stealer (a “Mini Shai-Hulud” variant whose source had been published openly). Wiz counted at least 32 releases (Socket put the figure higher), downloaded around 80,000 times a week; the payload ran via npm preinstall hooks and harvested GitHub, AWS, GCP, Azure, Kubernetes and Vault secrets. The intrusion traced to one compromised employee GitHub account. Red Hat pulled the packages and said they were internal tools never intended for customers, with no known production impact. Status: Confirmed. Editorial severity: 7/10. Takeaway: installation is code execution — pin and hash-verify dependencies, pull from a vetted internal registry, and scope CI tokens tightly.
• LiteLLM command injection (CVE-2026-42271). A command-injection flaw in the widely used LiteLLM gateway was added to CISA KEV as actively exploited — a reminder that the AI toolchain itself is now a target, not just a tool. Status: Confirmed (KEV). Editorial severity: 6/10. Takeaway: treat AI gateways and proxies as sensitive, credential-adjacent infrastructure and keep them patched and egress-restricted.

Confirmed: major data breaches

iRhythm Technologies — roughly 8.18 million people. Status: Confirmed (count per state Attorney-General filings; company has not confirmed the total). Editorial severity: 8/10. The heart-monitoring maker identified unauthorised access to business applications on a third-party platform on 8 June, was contacted by an extortionist on 9 June, and disclosed a material incident in an SEC 8-K on 18 June; it began notifying individuals in October. The Oregon Attorney General lists 8,179,772 affected; stolen data includes names, contact details, dates of birth, insurance numbers, device serial numbers and dates of service. Takeaway: health and device data carries long-tail identity risk — enforce MFA on any data-transfer platform and treat third-party app hosting as in-scope.

Novo Nordisk — confirmed intrusion, terabyte claim. Status: Confirmed intrusion; data volume claimed. Editorial severity: 7/10. The pharmaceutical company disclosed on 11 June that attackers reached a limited number of internal systems holding some personal data. A group calling itself FulcrumSec claimed to have taken more than a terabyte and demanded $25 million; a second group, TheUSERS007, reportedly demanded $50 million. Novo did not pay and said the clinical-trial data involved was pseudonymised, so it does not consider participants identifiable. Takeaway: do not confirm an attacker’s scope figures in your own comms until you have verified them; pseudonymisation of research data demonstrably limits the blast radius.

• Tata Electronics — ~630 GB claimed (Apple, Tesla files). The Apple and Tesla supplier confirmed a cybersecurity incident after World Leaks posted, on 12 June, more than 200,000 alleged files totalling about 630 GB, said to include iPhone 18 Pro engineering detail and Tesla component documents. Reuters reviewers could not immediately authenticate the files; Apple said it was investigating; Tata said operations were not affected. Status: Confirmed intrusion; data claims unverified. Editorial severity: 7/10. Takeaway: your suppliers’ breaches become your intellectual-property leaks — extend monitoring and contractual security requirements down the manufacturing chain.
• Eastman Kodak — 2.2 million records claimed. Kodak confirmed that an unauthorised third party temporarily accessed a limited amount of company data; ShinyHunters claimed 2.2 million records with customer PII and set a 18 June leak deadline, but published no samples to support the figure. Status: Confirmed (limited) intrusion; 2.2M claimed (unverified). Editorial severity: 5/10. Takeaway: a confirmed-but-limited breach and a seven-figure attacker claim are two different facts — hold the line between them in your disclosures.
• GrayRobinson (US law firm) — 65,000-plus people. The firm was reported to have had personal and health information stolen affecting more than 65,000 individuals. Status: Reported. Editorial severity: 5/10. Takeaway: professional-services firms concentrate sensitive client data — treat them as high-value targets and vendors.

Claimed, developing and nation-state

• Iranian banks — mid-June card and banking disruption. Several Iranian lenders, including Bank Melli, Bank Saderat, Bank Tejarat and the Export Development Bank of Iran, saw card payments, online banking and ATMs disrupted in a single week from around 14 June; officials said no customer data was compromised. Attribution is unresolved: the pro-Israel group Predatory Sparrow was widely named, but a cybersecurity academic said the attack did not fit that group’s pattern, and an anti-government group called Black Wolves claimed responsibility. Status: Disruption confirmed; attribution disputed. Editorial severity: 6/10. Takeaway: resist the urge to assign a known actor from pattern alone — early attribution is often wrong.
• Mastra AI framework — reported North Korea link. Reporting linked a supply-chain compromise of the Mastra AI agent framework to North Korea-aligned actors. Status: Reported (secondary sourcing; treat attribution cautiously). Editorial severity: 5/10. Takeaway: AI frameworks are now in the same threat model as any other dependency.
• “24 billion credentials” compilation. A researcher warning circulated about a compiled set of tens of billions of credentials. These compilations are almost always recycled from prior breaches and credential-stuffing lists, not evidence of a new mass breach. Status: Reported / recycled. Editorial severity: 4/10. Takeaway: the defence is unchanged — phishing-resistant MFA and no password reuse neutralise recycled credentials.

What to do this week

• Patch internet-facing enterprise software first. Mitigate and patch Oracle PeopleSoft (CVE-2026-35273), apply June’s Microsoft updates including the exploited Defender fix, and work the CISA KEV additions (SimpleHelp, Ivanti Sentry, Check Point, Splunk, Cisco, Ubiquiti) by your own exposure.
• Harden the software supply chain. Pin and hash-verify npm and PyPI dependencies, pull from a vetted internal registry, scope CI/CD tokens, and audit for the Red Hat package worm and the LiteLLM flaw.
• Lock down identities and OAuth tokens — the recurring entry point this month — with least privilege, short expiry and regular re-consent.
• Treat SaaS and suppliers as in-scope vendor risk. Inventory what each platform holds and score it with our Vendor Risk Scorecard, and understand how a leak spreads using our primer on data leakage.
• Build in layers so one compromised credential is not game over — see defense in depth — and put a defensible figure on exposure with the Breach Cost calculator and our explainer on what a data breach actually costs.

Frequently asked questions

Biggest attacks of June 2026? The Oracle PeopleSoft zero-day (CVE-2026-35273) that ShinyHunters used against 100-plus organisations, Microsoft’s record ~206-CVE Patch Tuesday with an exploited Defender flaw, and the Red Hat npm supply-chain worm; plus breaches at iRhythm, Novo Nordisk, Tata Electronics and Kodak.

How were the severity scores decided? A subjective PlayCISO reading out of ten — not CVSS or any standard. Use it as a relative signal only.

Were the threat actors’ numbers real? Treat them as unverified marketing: ShinyHunters’ 2.2M Kodak figure, World Leaks’ 630 GB Tata figure and FulcrumSec’s terabyte Novo Nordisk claim were not confirmed by the companies.

Was AI involved? The AI toolchain was mostly a target — an exploited LiteLLM command-injection flaw and a reported Mastra framework compromise — rather than an autonomous attacker. Treat “the AI did the hacking” framing with caution.

What to do first? Patch PeopleSoft and June’s Microsoft and KEV items, harden npm/PyPI dependencies, lock down OAuth app identities, and treat SaaS and suppliers as in-scope vendor risk. For the next instalment in this series, see our October 2026 roundup.